List your company

Image Captioning on DOCCI 500 (test)

64.31CAPTURESFT + Ours
50.176453.845757.51561.1843Mar 2, 2026

Evaluation results

MethodLinks
SFT + Oursbase_model=Qwen2-VL-7B SFT, method=CIM2026.0364.3145.1624.88——73.87——58.6836.32
7B-OursBase Model=Qwen2.5-VL, Method=CIM2026.0363.4639.822.7279.2866.9771.8873.5250.4559.0834.7
CIMBase Model=Qwen2.5-VL-7B2026.0363.46——79.2866.9771.8873.5250.4559.0834.7
7B-OursBase Model=Qwen2-VL, Method=CIM2026.0363.1237.2121.5381.9964.6171.4374.4950.1959.1832.12
CIMBase Model=Qwen2-VL-7B2026.0363.12——81.9964.6171.4374.4950.1959.1832.12
8B-OursBase Model=InternVL3, Method=CIM2026.0362.1828.2518.5482.7362.8670.4776.2249.5859.2630.39
CIMBase Model=InternVL3-8B2026.0362.18——82.7362.8670.4776.2249.5859.2630.39
72B-BaseBase Model=Qwen2.5-VL2026.0361.0933.7219.0182.1762.9270.3572.3746.1855.4630.71
76B-BaseBase Model=InternVL22026.0361.0938.6921.2181.4264.7771.1970.946.855.5729.67
Claude-3.72026.0361.0241.3620.17——69.48——55.5728.78
8B-OursBase Model=InternVL2, Method=CIM2026.0360.8232.0119.4982.1660.3568.6674.1647.5156.9826.36
CIMBase Model=InternVL2-8B2026.0360.82——82.1660.3568.6674.1647.5156.9826.36
8B-OursBase Model=InternVL2.5, Method=CIM2026.0360.7530.8917.8482.1762.2869.9972.9646.6456.0728.9
CIMBase Model=InternVL2.5-8B2026.0360.75——82.1762.2869.9972.9646.6456.0728.9
Gemini-1.52026.0360.3424.716.25——68.17——55.6928.9
72B-BaseBase Model=Qwen2-VL2026.0359.5630.7917.5784.0260.2169.1471.6344.3153.9227.21
78B-BaseBase Model=InternVL2.52026.0359.3723.615.7985.7859.1268.875.1245.555.7426.92
8B-BaseBase Model=InternVL22026.0358.8331.4917.9781.4959.5467.7270.8444.2453.5122.65
BaseBase Model=InternVL2-8B2026.0358.83——81.4959.5467.7270.8444.2453.5122.65
8B-BaseBase Model=InternVL2.52026.0358.6424.2315.2984.1158.0767.6273.4844.7954.6924.63
BaseBase Model=InternVL2.5-8B2026.0358.64——84.1158.0767.6273.4844.7954.6924.63
78B-BaseBase Model=InternVL32026.0358.5218.3213.8986.1858.2368.375.6944.3355.0127.85
7B-BaseBase Model=Qwen2-VL2026.0357.9629.3916.5983.6956.7966.4769.9643.2752.6517.57
BaseBase Model=Qwen2-VL-7B2026.0357.96——83.6956.7966.4769.9643.2752.6517.57
Qwen2-VL-7B2026.0357.9629.3916.59——66.47——52.6517.57
ShareCaptioner2026.0357.939.0923.05——66.05——52.2719.47
CIMBase Model=LLaVA1.5-7B2026.0357.1——83.1252.7863.3872.0647.4256.2819.87
8B-BaseBase Model=InternVL32026.0356.9512.5711.6685.9855.3966.0874.7543.1953.7225.11
BaseBase Model=InternVL3-8B2026.0356.95——85.9855.3966.0874.7543.1953.7225.11
7B-BaseBase Model=Qwen2.5-VL2026.0355.8922.6814.6784.6454.9665.0672.1542.1352.2724.35
BaseBase Model=Qwen2.5-VL-7B2026.0355.89——84.6454.9665.0672.1542.1352.2724.35
BaseBase Model=LLaVA1.5-7B2026.0350.72——81.1348.3759.4164.1239.7848.019.19